摘要
在以运动参数误差为状态量、视觉导航与惯导导航相对运动参数差为观测量的传统惯性/视觉组合导航方法中,为解决相对运动参数同时与前后两个时刻状态相关的问题,采用将前一时刻位置和姿态误差增广到状态量中的方法,并且假设增广的状态量为常值,导致状态模型中引入了较大的误差.基于真实位置、姿态建立观测量误差模型,导致观测量同时与前后两个时刻的状态相关.本文以惯导误差方程为状态模型,采用四元数差形式的相对运动参数差作为观测量,基于上一时刻组合导航位置、姿态估计值建立观测量误差模型,实现了状态的增广,并使得量测信息仅与当前时刻的位置误差和平台失准角相关,克服了状态模型误差较大的问题.月面仿真和地面模拟实验均表明,该方法能够达到较高的位置和姿态估计精度.
In traditional INS/VNS integrated navigation,the motion errors are usually used as the state vector,and relative motion errors between the inertial and vision navigation are used as the measurement.Since the relative motion is related to both the last and current states,traditional methods augment the position and attitude errors at the last time to the state vector to build the measurement model.The augmented states are considered as constant,and it generates new errors into the state model.Meanwhile,the measurement errors are analyzed based on ideal positions and attitudes at both the last and current time,which results in the measurement relationship with both the last and the current states.In this paper,a new INS/VNS model uses INS error equation as the state model,relative motion errors as the measurement,and attitude errors are described as quaternion error in the measurement model.The analyses of measurement errors are based on the integrated navigation estimation positions and attitudes at the last time,hence it does not need to augment the state,so the measurement only relates to the current state.The lunar surface simulation and experiment on the ground both show that the represented INS/VNS method can achieve high position and attitude estimation accuracy.
出处
《空间科学学报》
CAS
CSCD
北大核心
2015年第6期721-729,共9页
Chinese Journal of Space Science
基金
国家自然科学基金项目资助(61233005)
关键词
惯导误差方程
视觉导航
组合导航
巡视器
卡尔曼滤波
INS error equation
Visual navigation
Integrated navigation
Exploration rover.Kalman filter
